Fatal Error

[Rchapin]

Hey guys, back at Zoic and still using Phoenix. Really hoping we can finally fix some of these bugs. Biggest one now is:

When emitting fluid from particles with a limited lifespan, Maya will crash when the last particle dies.

Bob

[Svetlozar Draganov]

Hello Bob,

Can you send us or attach sample scene which reproduce the issue you are facing.
Please let us also know the exact versions/service packs /revision number(nightly build) of Phoenix FD , Maya and OS.

Thank you very much in advance.

[Rchapin]

For some reason, the forum will not let me attach the file. Perhaps because it's maya binary. But the bug is easily repeatable and I tested it under multiple versions of phoenix (2.2 & 2.1.1)
Currently using Maya 2014, but I've had this happen in previous versions.
Using Windows 7 SP1

create a particle emitter inside a fluid container
give the particle a lifespan
have the particles emit fluid

everything is fine at this point
took a little detective work to find out that the problem is caching the particles.

cache the particles under dynamics/solvers/create particle disk cache
run the fluid sim again

in the 2.1.1 build of Phoenix, the viewport completely disappears.
in the 2.2 version, fatal error.

Bob

[Ivaylo Katev]

the forum allows attaching only zip and jpg files, just zip it

[Rchapin]

It would almost be easier to just follow the 2-3 steps above, but here ya go...

[Rchapin]

Also don't know if it's ever been mentioned, but adjusting values for offset and scale under rendering/fire or density while simulation is running will kill any custom entries in the respective graph editor.

[Svetlozar Draganov]

I've just run a quick test on the scene you sent us following the steps from your previous post but I couldn't reproduce the issue.
Would you please update to the latest available Nightly Build of Phoenix FD and run the same test again?

Thank you very much in advance.

[ivaylo.ivanov]

This is a known issue and is a bug in Maya with some cache methods. I suggest you to use nParticles with their cache, it's the only one I think is not bugged, also you can specify a cache interval. The crash is when you set the time between two cached frames and some particles die between them (you can reproduce this without Phoenix, just set the frame to a fractional number). Phoenix do this during the simulation if the Steps per frame is different then 1.0 (so you can use this to prevent the crash)

[ivaylo.ivanov]

Also don't know if it's ever been mentioned, but adjusting values for offset and scale under rendering/fire or density while simulation is running will kill any custom entries in the respective graph editor.

Yep, it's a bit of a fight with this custom control and there is definitely room for improvement. We may need to move to using directly QT though, since doing a graph control with what Maya provides turned out be a big pain...

[Rchapin]

I see the confusion. There doesn't seem to be a problem with nparticles, I'm talking about regular particles.

Nparticles has actually been buggy in Maya and we went back to using regular particles in sims because they interpolate correctly between frames.

You can see this during a Phoenix sim. Regular particles will update their position on a subframe basis regardless of the number of steps per frame.

Not sure how you can crash Maya without Phoenix by setting a frame to a fractional number:

"The crash is when you set the time between two cached frames and some particles die between them (you can reproduce this without Phoenix, just set the frame to a fractional number)."

[vlado]

This is a known issue and is a bug in Maya with some cache methods. I suggest you to use nParticles with their cache, it's the only one I think is not bugged, also you can specify a cache interval.

Actually the nParticles are bugged. The normal particles interpolate fine for subframes.

Best regards,
Vlado

[Rchapin]

Yes, that's my point.

Still doesn't address my initial post that I get a fatal error whenever my last emitting particle (regular particle) dies.

And yes, it looks like it's OK if I only use 1 step per frame, but that's not really an option for the majority of our work.

[ivaylo.ivanov]

Well, if the nParticles are bugged, there isn't working particle cache in Maya . The old particles are also bugged.
Open your scene -> cache the particles -> delete Phoenix if you want -> enter 13.5 for current time - > crash
The issue with interpolation is totally different Maya bug. This one is crash that happens when particles are killed.

[ivaylo.ivanov]

My idea was to use nParticles and set the cache step to correspond to the step that Phoenix is doing (i.e. every 0.5 frames in the example). This should avoid both bugs I guess.

[Rchapin]

I've tried reproducing your bug with no luck - caching regular particles and setting time to a subframe (13.5) with no crash. And I've never seen anything like that in 15 years of using Maya (even though there's plenty of other bugs). What system, version number, build are you using?

So far, I've tried it on everything from Maya 8.5 to 2014 x64 (Windows) with and without Phoenix. The only real problem is Maya 2013 where regular particles wouldn't cache at all. So it's possible they broke something in the regular particle build.